Backup Hash files

[yiminghu]

Hi,

We're are in the middle process of moving to produciton. One of task is to prepare disaster recovery plan. I am wondering how to backup and hash files in case there is a problem with server? Is it enough to backup the folder which stores hash file? Or we have to use different mechanism?

thanks a lot,

Yiming

[chulett]

If you've created them using the 'pathed' option, then yes, backing up (in some form) the entire contents of their folder is sufficient.

[ray.wurlod]

The usual precautions about backing up databases apply. Make sure that they're not in use before starting the backup.

You can detect which dynamic hashed files are in use via the command analyze.shm -d (the executable is in the bin directory in the DataStage Engine directory). smat (shared memory analysis tool) is a synonym for analyze.shm

[olgc]

[quote="chulett"]If you've created them using the 'pathed' option, then yes, backing up (in some form) the entire contents of their folder is sufficient.[/quote]

But How can it be backed up if the Account Name is used with empty field?

Thanks,

[kcbland]

If you've created them using the 'pathed' option, then yes, backing up (in some form) the entire contents of their folder is sufficient.

But How can it be backed up if the Account Name is used with empty field?

Thanks,

Than your only choice is to backup the project, and since you can't separate job designs from hash file designs easily, it's an all-or-nothing restore. That would be why you shouldn't let data go into the project directory. It's like storing Word or Excel documents in the System32 folder.

[bdstage]

Will the indexes are also preserved if we backup the hash file by simply copying it from its source directory?

Thanks in advance for the reply.

Thanks,
Pavan.

[kcbland]

Yes, but externally pathed hashed files don't have a pointer (synonym) in the project, so you can copy them around to anywhere easily. Project-defaulted hashed files assume the file is in the project directory.

[ray.wurlod]

Will the indexes are also preserved if we backup the hash file by simply copying it from its source directory?

Thanks in advance for the reply.

Thanks,
Pavan.

Indexes will NOT be preserved. By default a hashed file does not have any indexes, so that's not a problem. If you have created index(es) you will need to back up the I_hashedfile directory for each. You should also have backed up the D_hashedfile file (the dictionary portion) for each.

[kcbland]

My apologies, I wasn't clear enough. I didn't clarify my thoughts that the poster doesn't have to worry about indexes since hashed files don't have them unless you specifically create them. I mentally figured that they mean "primary key index" which of course doesn't exist on a hashed file. Secondaries are other issues as Ray has noted.

[bdstage]

Thanks Ray and Kenneth for your valuable inputs. It clearly helped me in understanding the hash file back up process and for good DR.

Thanks a lot.